Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jan 18.
Published in final edited form as: Prev Sci. 2026 Jan 7;27(2):208–220. doi: 10.1007/s11121-025-01873-0

Family engagement in an early childhood preventive parenting program: Innovative methods for examining sociodemographic, psychosocial, and contextual predictors

Yu Chen 1,*, Caitlin F Canfield 1, Eric D Finegood 2, Juliana Gutierrez 1, Alyssa Milton 2, Kyrstin Loney 2, Lauren K O’Connell 3, Alan Mendelsohn 1
PMCID: PMC12812042  NIHMSID: NIHMS2135651  PMID: 41499075

Abstract

Despite well-studied benefits of preventive parenting programs for early child development, various real-world barriers may impede families from engaging in those programs. The current study aims to provide new insights into family engagement by examining enrollment, retention, and involvement and their predictors in an evidence-based universal pediatric primary care parenting program for families with young children. Data (n = 204) were from an ongoing longitudinal randomized controlled trial of PlayReadVIP in Flint, MI. Families (66% Black, 35% White, low socioeconomic status) had high enrollment and retention in the program. As exploratory analyses, random forest models, a machine learning method, identified a multitude of sociodemographic, psychosocial, and contextual predictors of retention and involvement in PlayReadVIP across the first 9 months. As confirmatory analyses, multiple regressions showed that COVID-19 significantly hindered retention (odds ratio = .04; b = −.30) and involvement (b = −.31) and that higher parenting self-efficacy was associated with lower retention (odds ratio = .76). Furthermore, the association between COVID-19 and family engagement was moderated by household income, suggesting that families with the highest economic risks were less likely to attend and be actively involved in sessions during the pandemic. This study addresses important research gaps by focusing on multiple aspects of family engagement in a pediatric program during infancy, assessing whether experiencing contextual adversity hinders or motivates engagement, and employing a machine learning method. These findings have crucial implications for designing and implementing early childhood prevention parenting programs to more effectively engage families with higher needs.

Keywords: Preventive parenting program, Family Engagement, COVID-19, Pediatric, Random Forest

Preventive parenting programs (PPPs) during early childhood have shown significant effects on improving positive parenting behaviors and child developmental outcomes for families experiencing various sociodemographic and psychosocial risks (Jeong et al., 2021; Sandler et al., 2011). However, systematic reviews and meta-analyses indicate that those effects are only small to moderate (Zhang et al., 2021). A key determining factor for effectiveness is the extent to which families are engaged in the program, and higher engagement has been associated with stronger program impacts on parents and children (Brody et al., 2006; Burney et al., 2024). One of the particular challenges of family engagement in PPPs is that the risk factors that tend to be associated with lower engagement are also the same ones that are frequently associated with reduced positive parenting and lower developmental skills (Robinson et al., 2016). Therefore, programs are often not utilized by families who, based on their existing risks, might need them the most. When implemented in real-world settings, early childhood PPPs often fail to achieve the same level of family engagement as in efficacy trials, with 35-80% of the eligible families enrolling in the program (e.g., participating in at least one session) and the enrolled families attending 40-70% of the sessions (Baker et al., 2011; Garvey et al., 2006; Hackworth et al., 2018; Heinrichs et al., 2005; O’Brien et al., 2012). Understanding the predictors of family engagement is an important step towards more effectively enrolling and retaining families in PPPs, especially those facing multiple disparities who may benefit the most from such support.

Definition and Predictors of Family Engagement

In the field of prevention science, family engagement consists of multiple aspects, including recruitment, enrollment, attendance and retention across program sessions, participation and involvement in specific program components, and families’ satisfaction about the program and relationship with program staff (Eisner & Meidert, 2011; Hackworth et al.,2018). For this study, we focus on three aspects—enrollment, retention, and involvement—to comprehensively capture the behavioral dimensions of family engagement. Enrollment represents families attending at least one program session; retention refers to the number or proportion of sessions attended among those who had at least one session; involvement is defined as families actively practicing learned skills during a session (Baker et al., 2011; Hackworth et al., 2018; Haine-Schlagel & Walsh, 2015; Prado et al., 2006).

Family Factors

Sociodemographic characteristics, such as younger maternal age, single parenthood, a higher number of children, lower income and education, and racial and ethnic minority identity, have been associated with lower family engagement in parenting programs (Heinrichs et al., 2005; Mauricio, Mazza, et al., 2018; Reyno & McGrath, 2006; Whittaker & Cowley, 2012). For example, a behavioral parent training program for preschoolers found that parents with lower incomes and African American (compared to Caucasian) parents were less likely to enroll (Baker et al., 2011). Similarly, a study on a home visiting program for low-income pregnant women and parents of young children showed that younger, unmarried, unemployed, and African American parents were more likely to drop out of the program and completed fewer home visits (O’Brien et al., 2012). However, two studies demonstrated nonsignificant relations between socioeconomic status and family engagement ((Damashek et al., 2011; Katz et al, 2001), which may be attributed to their focus on high-risk families (i.e., mothers without prenatal care, and parents involved in child welfare cases).

Psychosocial factors, such as maternal depression, parenting stress, parenting self-efficacy, and social support, have also been associated with family engagement (Reyno & McGrath, 2006). A home-based child maltreatment prevention program showed that higher maternal depression was associated with higher likelihood of enrollment (Damashek et al., 2011), which may be explained by parents’ perception of the program as a source of social support (Staudt, 2007). However, a study on a community-based parenting program found that higher psychological distress and more stressful life events were associated with lower retention (Hackworth et al., 2018). The discrepancy in findings may be attributed to differences in family engagement measures—one study focused on enrollment and the other on retention. Previous research indicates that depressed parents may be more likely to enroll in a program but less likely to sustain their attendance over time compared to non-depressed parents (Mauricio et al., 2014). In addition, parenting self-efficacy may play a role such that parents who are less certain about their role are less likely to attend programs (Whittaker & Cowley, 2012). Although resilience has not typically been examined as a predictor of family engagement, it has been linked to positive parenting behaviors (Conger & Conger, 2002; Gavidia-Payne et al., 2015), and parents’ ability to adapt successfully in the context of adversity may help sustain their engagement during the COVID-19 pandemic (Prime et al., 2023), an important context for this study.

Contextual Factors Outside the Home

Research exploring the relation between contextual factors outside the home and family engagement in PPPs is scarce. Neighborhood conditions is one of the more commonly examined factors. Similar to findings on family SES, higher neighborhood SES has also been associated with higher enrollment in parenting programs (Hackworth et al., 2018). Moreover, more neighborhood material and social resources, such as availability of public transportation, physical order, and community networks and cohesion, have been linked to increased enrollment (Hackworth et al., 2018; Eisner & Meidert, 2011).

COVID-19 as an Important Context.

The pandemic introduced numerous stressors to families with young children, including lack of access to child care, loss of employment and income, social isolation, and exposure to viruses and illness, all leading to more challenges in parenting (Desimpelaere et al., 2023). On one hand, these challenges may motivate parents to engage in programs that support positive parenting and family functioning. On the other hand, heightened financial and psychosocial stress may hinder parents from participating in PPPs because it is “an extra thing” that they need to devote time and efforts to (Barnett et al., 2023). Furthermore, family characteristics may moderate the relation between COVID-19 and engagement in PPPs. Compared to more advantaged families, families from under-resourced backgrounds, who tend to be impacted more substantially by the pandemic (Oberg et al., 2022), may be less likely to engage in PPPs during COVID-19 due to exacerbated difficulties. Conversely, those families may be more likely to participate due to increased needs. To our knowledge, no prior research has examined COVID-19 as a predictor of family engagement in early childhood PPPs. This study is the first to explore the association between COVID-19 and family engagement and whether family factors strengthen or mitigate this association.

Design and Impact of PlayReadVIP

PlayReadVIP, an early childhood PPP, is designed to promote positive parenting and school readiness through reading and play. Each program session takes 20-25 minutes, during which a trained, bachelor’s level “PlayReadVIP Coach” provides the family with a developmentally appropriate toy/book, video records the parent and child interacting with their new toy/book, and reviews the video with parents in real time, highlighting responsive interactions and discussing ways to expand positive parenting and foster developmental skills at home (Figure 1). Sessions are available to families approximately every 3 months from birth to age 3 years and typically take place in a clinic/hospital when families come for their well-child visits. PlayReadVIP coaches complete a 3-day training course, including detailed review of background material, and instruction regarding modeling and culturally-sensitive practice. Certification is attained through a three-stage process: 1) co-leading sessions with experienced coaches; 2) supervisor-observed independent sessions; and 3) competency assessment using a rubric of observed skills and a written exam. Fidelity is maintained through ongoing supervision, age-specific checklists to facilitate feedback, observation to monitor quality, and weekly supervisor meetings.

Figure 1.

Figure 1

Open in a new tab

PlayReadVIP Session Flowchart

Through three prior randomized controlled trial (RCT) studies, PlayReadVIP has shown significant effects on a range of parent and child outcomes, including parent psychological well-being, positive parenting behaviors, responsive parent-child interactions, and children’s school readiness, during infancy and toddlerhood and sustained through age 4.5 years (Cates et al., 2018; Mendelsohn et al., 2018; Miller et al., 2023, 2024).

Family Engagement in PlayReadVIP

Previous RCT studies of PlayReadVIP have shown that 95-99% of families attend at least one session, while close to 90% of those families attend half or more of the sessions during the first six months (Canfield et al., 2023; Miller et al., 2020) and two thirds do so across the first three years (Cates et al., 2018; Weisleder et al., 2016). Additionally, less-educated mothers and mothers reporting lower levels of parenting self-efficacy attended more PlayReadVIP sessions during the first 6 months (Miller et al., 2020). This may be attributed to their stronger need for resources and support and highlights the program’s ability to engage families who are traditionally hard to reach but need such services the most. PlayReadVIP has been implemented beyond RCTs in socioeconomically and culturally diverse communities and has shown effectiveness in those contexts (Piccolo et al., 2023).

A few features of PlayReadVIP are particularly relevant for studying family engagement. First, by aligning the sessions with pediatric well-visits (i.e., families can complete the session while waiting for their well-visit appointment or right after it), the program potentially reduces barriers to engagement due to lack of time and stigma about receiving intervention (Miller et al., 2020). Second, PlayReadVIP cultivates a long-term positive relationship between the coach and the family through one-on-one sessions and strength-based approaches that empower parents in their role. Third, remote delivery of PlayReadVIP was rolled out after the COVID-19 lockdown to continue engaging families safely and alleviate transportation difficulties, another major barrier to engagement (Whittaker & Cowley, 2012). This method has been adopted by several other parenting programs during the pandemic and has shown comparable effects to in-person delivery as well as high acceptance among parents (Solís-Cordero et al., 2022).

Random Forest as an Innovative Exploratory Method

Random forest (RF), a machine learning method, offers several advantages over linear regression when exploring predictors of family engagement. Unlike linear regression, which is constrained by the structure of the data (e.g., linear relation, multicollinearity, skewness, etc.), RF is a non-parametric approach that can account for complex, non-linear relations, including interactions among predictors as well as multicollinearity, when making predictions (Loef et al., 2022; Rothacher & Strobl, 2024). Additionally, RF is robust to outliers and can identify important predictors even amongst many irrelevant ones. As a result, RF tends to outperform traditional statistical methods in terms of predictive accuracy and reduces the risk of overfitting (Fife & D’Onofrio, 2023; Rothacher & Strobl, 2024). In cases where many predictors could be associated with family engagement, narrowing down the selection based on predictor importance reduces overfitting and Type I errors that tend to result from conducting a number of regression models. By ranking predictors based on their importance, RF serves as a valuable tool for exploratory analysis, which can guide further analysis in examining the most influential predictors in regression models. This complementary approach leverages the strengths of RF while allowing regression models to provide interpretable estimates and hypothesis testing for the key predictors identified.

The Current Study

Using data from an ongoing RCT of PlayReadVIP in Flint, MI, this study focuses on two research questions: 1) which family factors and contextual factors outside the home are associated with family engagement—enrollment, retention, and involvement—in PlayReadVIP during the first 9 months; and 2) whether household income, parent education, and parenting self-efficacy moderate the association between COVID-19 and family engagement. In line with most studies, we hypothesize that sociodemographic risks would be associated with lower family engagement while acknowledging that an opposite relation is possible. We also hypothesize that psychosocial risks may be associated with higher engagement considering the program’s emphasis on promoting parenting self-efficacy and maintaining a supportive relationship with parents. Because families in Flint, MI have been facing entrenched economic and racial disparities at the collusion of two public health crises—the Flint Water Crisis and COVID-19 (Ezell et al., 2021; Sadler & Highsmith, 2016), we expect worse neighborhood conditions and experience of the pandemic to be associated with lower levels of family engagement. Lastly, based on previous findings that COVID-19 disproportionally affects low-income families and that parent education and parenting self-efficacy are related to attendance in PlayReadVIP, we explore whether these family factors moderate the association between COVID-19 and family engagement. Although a large body of literature has focused on understanding the patterns of family engagement and its precedents, few studies have examined engagement in PPPs starting from birth or PPPs designed to mitigate barriers for engagement by leveraging healthcare partnerships, strength-based approaches, and remote delivery. Furthermore, we intend to provide a practical example of using RF to obtain more robust evidence on family engagement.

Method

Participants

Families were from an ongoing longitudinal RCT of PlayReadVIP. Children were eligible to participate in the RCT if they were born at or after 32 weeks; weighed 1,500 grams or more at birth; were singleton birth; had no significant developmental or physical impairments; and had no significant neonatal or medical complications. Parents were eligible if they spoke English; had custody of the child; did not have a significant communication impairment; and planned to continue receiving pediatric care at the Hurley Children’s Clinic in Flint, MI. In addition, families were eligible to enroll within three months of their child’s first well-child visit. A total of 486 families were enrolled, and 408 families were randomized into the intervention or control (care-as-usual) condition. This study focused on the intervention families (n = 204).

Table 1 shows the sociodemographic characteristics of the sample. Close to half (46.6%) of the parents were first-time parents and two-thirds (65.7%) self-identified as Black/African American. Half of the households reported annual incomes of $30,000 or lower, and parents on average had 12.6 years of education. Most (65%) of the families were living in Flint and the rest were living in nearby areas while accessing healthcare in Flint.

Table 1.

Sample Characteristics and Descriptive Statistics of Baseline Predictors and Family Engagement Variables in PlayReadVIP (n = 204)

Sample Characteristics and Predictors % (n)/M (SD) Missing % (n) Range
First-time parent 46.6% (95) 3.4% (7) -
Parent race (including multiracial) 0% (0) -
 Black/African American 65.7% (134)
 White 34.8% (71)
 Other (e.g., American Indian, Pacific Islander) 4.9% (10)
Parent is Hispanic/Latinx (any race) 2.5% (5) 1.5% (3) -
Household annual income 14.2% (29) -
 $30,000 or lower 51.0% (104)
 $30,001 - $60,000 25.0% (51)
 More than $60,000 9.8% (20)
Parent education (years) 12.60 (2.32) 2.0% (4) 2.5-23
Employed in the past 3 months 47.6% (97) 2.5% (5) -
Maternal age (years) 27.04 (5.5) 10.3% (21) 17.60-44.70
Married/cohabiting 54.9% (112) 2.0% (4) -
Living in Flint, MI 65.2% (133) 0% (0) -
Having material hardship (binary) 39.2% (80) 11.8% (24) -
Parent psychological distress 27.19 (6.09) 14.2% (29) 18.63-55.35
Parenting self-efficacy 18.01 (2.32) 13.7% (28) 11-20
Emotional support 4.21 (0.80) 10.3% (21) 1.71-5
Resilience 4.12 (1.09) 10.8% (22) 1-5
Ever received family service (binary) 42.7% (87) 11.8% (24) -
Had child care support (binary) 69.6% (142) 10.3% (21) -
Enrolled during COVID-19 (binary) 33.8% (69) 0% (0) -
Eligible PlayReadVIP sessions during COVID-19 (%) 31% (.35) 0% (0) 0-1
Neighborhood order 9.99 (3.49) 12.7% (26) .02-16.85
Neighborhood crime and safety −.20 (2.62) 11.3% (21) −7.32-3.58
Family Engagement in PlayReadVIP % (n)/M (SD) Analytic Sample Range
Enrollment (attended at least 1 session) 93.6% (191) n = 204 -
Retention
 High (3-5 sessions) 77% (134) n = 173 -
 Low (1-2 sessions) 23% (39) n = 173 -
 Percent eligible sessions attended 73% (.28) n = 191 0-1
Involvement
 High (3-5 videos) 49.7% (86) n = 173 -
 Medium (1-2 videos) 42.8% (74) n = 173 -
 Low (0 video) 7.5% (13) n = 173 -
 Percent eligible videos/reviews completed 48% (.29) n = 191 0-1
Open in a new tab

Procedure

Families were typically recruited and consented in person at their first well-child visit at the clinic. During the COVID-19 lockdown, families were first asked by their doctor or nurse who was seeing them at the clinic if they were willing to be contacted for the RCT, and those who agreed were then approached by researchers via phone. Families who agreed to participate completed electronic consent on the phone with a researcher. Families randomly assigned to the intervention condition were offered up to 14 PlayReadVIP sessions from birth to 36 months. The content and structure of the sessions are described above. Sessions took place in person at the time of well-child pediatric visits or remotely through Zoom, which was made available to parents starting May 2020 in response to the COVID-19 social distancing policy and remained an option after the pandemic lockdown. Remote sessions included the same content and structure as in-person sessions. Through interviews with a researcher, parents completed a baseline survey on sociodemographic characteristics, parent well-being, emotional support, and neighborhood conditions at the time of enrollment from August 2019 to October 2023.

Measures

Family Engagement in PlayReadVIP

This study examined family engagement across the first five sessions (1-, 2-, 4-, 6-, and 9-month sessions) prior to the first follow-up research assessment. Families were eligible to attend each session within an age window: birth – child age 1.59 months for the 1-month session, 1.6 – 2.99 months for the 2-month session, 3 – 4.99 months for the 4-month session, 5 – 7.99 months for the 6-month session, and 8 – 10.99 months for the 9-month session. In addition, when families had enough time, they were given the opportunity to make up for the previously missed session within the current one. Make-up sessions were intended to ensure that families received as many relevant parenting and child development resources as possible. However, make-up sessions did not include video recording and review due to time constraints.

The PlayReadVIP coach recorded if a family fully or partially completed or missed each session. Partial sessions usually did not include the video recording and review component, but the coach and the parent still discussed current developmental milestones, generated ideas for using the book or toy, and set goals for home. Partial sessions occurred due to various reasons, including lack of time, the baby being fussy or sleepy, discomfort being video recorded, and make-up sessions. From a program delivery perspective, partial sessions were included for enrollment and retention, as families still made an effort to attend the current session or make up for a previous one, and still received information on parenting and child development.

Enrollment was dichotomized based on the total number of PlayReadVIP sessions each family attended, including fully and partially completed sessions. Families who attended at least one session were scored 1. Retention was measured among families who initiated enrollment. It was included as a binary variable (i.e., high retention refers to fully or partially completing three or more out of the five sessions and was coded as 1) and a percentage variable (i.e., number of fully and partially completed sessions divided by number of eligible sessions). Eighteen families consented for the RCT when the child was older and thus were not eligible for all five sessions. The binary variable excluded those 18 families. Similarly, involvement was measured as a categorical variable (i.e., high if families completed three or more video recordings and reviews, medium if one or two, and low if none) and a percentage variable.

Sociodemographic Factors

Sociodemographic characteristics at baseline were selected based on prior research. The following variables were binary: first-time parents, parents who were married or cohabiting with partner, living in Flint, employed in the past three months, self-identified as Black/African American, and having some material hardship. For material hardship, parents reported on four binary items (e.g., “During the past 12 months, has there been a time when your household had service turned off by the gas or electric company?”) from the Difficult Life Circumstances scale (Oxford et al., 2015). Scores were summed across the items and then dichotomized because 60% parents reported having no material hardship at baseline. Three continuous variables were also included: parents’ years of education, annual household income (reported as 11 categories and thus treated as continuous) and maternal age.

Psychosocial Factors

Parent psychological distress.

Four indicators reported by parents at baseline were used to construct parent psychological distress: anxiety, depression, stress, and post-traumatic stress disorder (PTSD). Anxiety, containing seven items (e.g., “In the past 7 days, I felt anxious.”), and depression, containing eight items (e.g., “In the past 7 days, I felt helpless.”), were measured using the PROMIS anxiety and depression subscales (Teresi et al., 2016a, 2016b). All items were on a 0-4 scale and summed scores were converted into t-scores for analysis. Stress was measured using the Perceived Stress Scale (Cohen et al., 1983), which included 10 items on a 0-4 scale (e.g., “In the last month, how often have you felt that you were unable to control the important things in your life?”). PTSD was measured using the Breslau short screening scale for DSM-IV PTSD (Breslau et al., 1999), and consisted of seven binary items (e.g., “In the past month, have you lost interest in activities that were once important or enjoyable?”). Scores for stress and PTSD were summed across items. Cronbach’s alphas for the four scales ranged between .80 and .93 in our sample, indicating acceptable to high internal consistency. Confirmatory factor analysis showed that the four indicators represented one latent construct (χ2(df =2) = 1.22, p = .54; CFI = 1.00; RMSEA = .00 and 90% CI = [.00, .13]). Higher average scores across the four indicators indicated more parent psychological distress.

Parenting self-efficacy.

At baseline, parents rated five items on a 0-4 scale using the Parenting Self-Agency Measure (Dumka et al., 1996), adapted for use with parents of newborns (i.e., changing to future tense). Items asked about the extent to which parents agreed with each statement (e.g., “I know I am going to do a good job as a parent.”). In our sample, the scale demonstrated approaching acceptable internal consistency (α = .67). Higher summed scores across items indicated higher parenting self-efficacy.

Emotional support.

Using the Social Resources subscale of the Resilience Scale for Adults (RSFA; Friborg et al., 2003), parents rated seven items on a 1-5 scale to indicate the extent to which they agreed with each statement (e.g., “Some close friends/family members are good at encouraging me.”). Higher average scores across items indicated more emotional support perceived by parents. The Cronbach’s alpha for this scale in our sample was .79.

Resilience.

Parents rated six items on the Personal Competence subscale of the RSFA (Friborg et al., 2003) to indicate their level of resilience in daily life (e.g., “When something unforeseen happens, I always find a solution.”). Higher average scores on a 1-5 scale represented more resilience (Cronbach’s alpha = .78).

Family service and child care support.

Two binary variables at baseline were included to indicate whether parents ever received services offered through Genesee County Health System since 2014 (e.g., connected with a Family Support Coordinator, who helped family connect with medical, educational, social, and other services needed) and whether parents had other adults in the home to provide child care support.

Contextual Factors Outside the Home

COVID-19.

This variable was operationalized in two ways. We calculated the percentage of sessions families were eligible for during the most intense COVID-19 period (March 2020 to March 2022) and also coded whether families enrolled into the RCT during that period.

Neighborhood conditions.

Parents responded to items adopted from the Speak to Your Health Community Survey measuring three aspects (Mathis et al., 2016): neighborhood cohesion (five items; e.g., “People in this neighborhood are willing to help their neighbors.”), physical environment (eight items; e.g., “My neighborhood is noisy.”), and crime and safety (three items; e.g., “How fearful are you about crime in your neighborhood?”). Items for each aspect were summed to create a composite score (individual items were standardized due to scale differences). Cronbach’s alphas for these composite variables ranged between .65 and .89. For analyses, we constructed an overall score to represent neighborhood order by averaging across the composite scores of neighborhood cohesion, physical environment, and crime and safety and also examined the composite score of crime and safety individually.

Analytic Plan

Exploratory Analyses Using Random Forest

We conducted RF using the Python package scikit-learn to identify important predictors of family engagement. Following guidelines from previous research (Loef et al., 2022; Rothacher & Strobl, 2024), the RF analyses included four main steps: 1) data preparation; 2) model specification; 3) prediction performance optimization; and 4) variable importance ranking. All predictors were included in the RF analyses, conducted separately for each family engagement measure. Detailed analytic methods for RF are included in the Supplementary Materials. The importance of each predictor was ranked based on the average change in prediction performance—accuracy or mean squared error (MSE)—for the corresponding family engagement measure when that predictor was permuted.

Confirmatory Analyses Using Regression Models

Based on the importance ranking, we selected the top five important predictors for each family engagement measure to include in the regression analyses (main effects). Linear and logistic regression models were used for continuous and binary/categorical family engagement measures, respectively. To test whether the association between COVID-19 and family engagement is moderated by family factors, we included interaction terms between COVID-19 and household income, parent education, and parenting self-efficacy in separate regressions models. We also examined simple slopes to probe the associations between COVID and family engagement at different levels of those moderators.

Results

Family Engagement in PlayReadVIP

As shown in Table 1, 191 (94%) families attended at least one session. However, because only 13 (6%) families did not attend any sessions, this subsample is too small to be predicted using RF or regression models and we thus did not conduct analyses on predicting enrollment. In addition, 77% of families showed high retention (i.e., attending three or more sessions) during the first nine months. Most families completed video recording and review at least once, with 43% demonstrating medium involvement (i.e., one or two times) and approximately 50% demonstrating high involvement (i.e., three or more times). See Table S2 in Supplemental materials for additional information on family engagement measures.

Results from RF Analyses

Predicting Retention

Table S1 in Supplemental Materials shows the top five important predictors for retention and involvement in RF analyses. For both measures of retention (i.e., high vs. low retention and percent eligible sessions attended), percent eligible sessions during COVID-19 was a high importance predictor, ranked second and first, respectively. Other important predictors included whether families ever received county services since 2014, parenting self-efficacy, maternal age, emotional support, household income, parent education, neighborhood crime and safety, and first-time parents. However, the latter four predictors, which were specifically important for predicting percent eligible sessions attended, were much less important compared to COVID-19 (e.g., decrease in MSE for COVID-19 was eight times larger than that for household income). This could suggest that COVID-19 may be the only key predictor for percent eligible sessions attended during the first nine months. In subsequent regression analyses for retention, we included all nine important predictors.

Predicting Involvement

The prediction performance for high vs. medium vs. low involvement was barely above chance (i.e., 55% accuracy), indicating that the predictors included could not yield a reliable prediction when involvement was categorized into three levels. This may be attributed to the considerably unbalanced distribution of the groups (i.e., only 8% of the sample showed low involvement; Table 1). Therefore, we did not conduct any subsequent regression analyses on this measure. For percent eligible video recording and review completed, percent eligible sessions during COVID-19 was again the most important predictor, followed by first-time parents, parenting self-efficacy, household income, and whether parents self-identified as Black/African American. Similar to the results for predicting percent eligible sessions attended, COVID-19 demonstrated substantially higher importance compared to the other predictors. All five predictors were included in the subsequent regression analysis for predicting percent eligible video recording and review completed.

Results from Regression Analyses

Predicting Retention

For high vs. low retention, the main effects model (Table 2) showed that for one unit increase in parenting self-efficacy, the odds of high retention decrease by 24% (OR = .76, 95% CI = [.59, .99]) and that for one unit increase in percent eligible sessions during COVID-19, the odds of high retention decrease by 96% (OR = .04, 95% CI = [.01, .18]). In other words, a higher level of parenting self-efficacy and a higher percentage of sessions during COVID-19 were both associated with a lower likelihood of retention. No significant interactions between family factors and COVID-19 were found.

Table 2.

Regression Models Predicting Retention and Involvement in PlayReadVIP

High vs. Low Retentiona % Eligible Sessions Attended (Retention) % Eligible Video Recording and Review Completed (Involvement)
Odds Ratio Robust SE Beta SE Beta SE Beta SE Beta SE
Parenting self-efficacy .76* .10 −.09 .01 −.07 .01 −.11 .01 −.07 .01
COVID-19 (% eligible sessions) .04*** .03 −.30*** .07 −.61*** .13 −.31*** .07 −.68*** .13
Household income 1.07 .11 .01 .01 −.10 .01 .03 .01 −.13 .01
First-time parent 1.45 .86 .12 .05 .12 .05 .11 .05 .10 .04
Family service 1.85 1.05 .09 .04 .10 .04 - - - -
Emotional support 1.41 .52 .16 .03 .14 .03 - - - -
Maternal age 1.02 .06 .03 .00 .05 .00 - - - -
Parent education 1.04 .14 .11 .01 .06 .01 - - - -
Neighborhood crime and safety .99 .11 −.05 .01 −.03 .01 - - - -
Black/African American - - - - - - −.14 .05 −.16* .05
COVID-19 x Household income - - - - .36* .02 - - .45** .02
Model fit X2(118) =109.48 p = .70 F(9, 130) = 3.21, p = .00, R2 adj = .13 F(10, 129) = 3.50, p = .00, R2 adj = .15 F(5, 140) = 5.73, p = .00, R2 adj = .14 F(6, 139) = 6.55, p = .00, R2 adj = .19
Open in a new tab

Note. Independent variables included the important predictors for each family engagement measure identified by the RF analyses.

COVID-19 = % eligible sessions during COVID-19 (March 2020 – March 2022). Family service = ever received services from Genesee County since 2014.

a

Low retention was the reference group coded as 0 and high retention was coded as 1.

*

p < .05.

**

p < .01.

***

p < .001.

For retention measured as percent eligible sessions attended, the main effects model (Table 2) showed that higher percent eligible sessions during COVID-19 was significantly associated with lower retention (β = −.30, p < .001). Furthermore, a significant interaction between COVID-19 and household income was found (β = .36, p < .05) and simple slopes indicated that COVID-19 was significantly associated with retention only at the 25th ($6,000 - $12,000 per year) and 50th ($18,000 - $24,000 per year) percentiles of household income (β = −.41, 95% CI = [−.60, −.22]; β = −.26, 95% CI = [−.39, −.12], respectively) but not at the 75th percentile ($36,000 - $42,000 per year). Interactions between COVID-19 and parent education and parenting self-efficacy were nonsignificant.

Predicting Involvement

The main effects model (Table 2) again showed a significant negative association between percent eligible sessions during COVID-19 and involvement (β = −.31, p < .001). Likewise, a significant interaction between COVID-19 and household income was found (β = .45, p < .01) and simple slopes revealed significant associations only at the 25th ($6,000 - $12,000 per year) and 50th ($24,000 - $30,000 per year) percentiles of household income (β = −.48, 95% CI = [−.67, −.29]; β = −.28, 95% CI = [−.42, −.14], respectively) but not at the 75th percentile ($36,000 - $42,000 per year). Interactions between COVID-19 and parent education and parenting self-efficacy were nonsignificant.

Discussion

This study examined sociodemographic, psychosocial, and contextual predictors of family engagement in an early childhood PPP, PlayReadVIP. It is one of the first studies to assess multiple aspects of family engagement in a pediatric-based universal program during infancy. Moreover, this study focused on families facing long-lasting adversities in Flint, MI, offering novel evidence on the extent to which these families participate in early childhood PPPs and its predictors. This study also employed RF as an exploratory machine learning method to robustly identify important predictors of family engagement among a multitude of potential factors, enabling subsequent confirmatory analyses to examine and interpret the relations between those predictors and family engagement.

Consistent with previous findings on PlayReadVIP (Canfield et al., 2023; Miller et al., 2020), the majority (94%) of families attended at least one session during the first nine months, highlighting the program’s ability to effectively enroll families from less advantaged backgrounds. Compared to the 90% high retention rate found in previous PlayReadVIP studies (Canfield et al., 2023; Miller et al., 2020), a lower rate (77%) was found in this study. It is possible that PlayReadVIP could not address certain barriers to engagement that resulted from COVID-19. Regardless, PlayReadVIP demonstrated impressively higher enrollment and retention than other early childhood PPPs (e.g., 35-80% enrollment rate and 40-70% retention rate; Baker et al., 2011; Garvey et al., 2006; Hackworth et al., 2018; Heinrichs et al., 2005; O’Brien et al., 2012). This may be explained by the program’s design to facilitate engagement by utilizing pediatric primary care, fostering coach-parent relationships, and offering remote delivery (Miller et al., 2020; Staudt, 2007; Whittaker & Cowley, 2012).

As identified by both RF and regression analyses, the percentage of sessions families were eligible for during the most intense period of COVID-19 was one of the most important predictors of retention and involvement. Importantly, COVID-19 was significantly associated with reduced engagement in PlayReadVIP only for lower-income families, especially those falling below the poverty line ($24,526 for a family of three in 2023; U.S. Census Bureau, 2024), whereas the association was not significant for families with incomes above the median household income in Flint ($36,194 in 2023; U.S. Census Bureau, 2024). Adding to previous research showing that low-income families are disproportionally affected by COVID-19 in regard to financial capacity, parent psychological well-being, and family functioning, our study suggests that these COVID-19-related stressors may further create barriers for families with the highest economic risks to participate in programs that could help them work through difficult situations. Nonetheless, the finding that engagement was not negatively affected by COVID-19 for families with fewer economic risks—although many are still considered low-income—indicates that PlayReadVIP is able to effectively engage families even during a challenging time like COVID-19. A promising strategy to expand the program’s reach to the most vulnerable families could be combining with other services aiming to improve parents’ livelihood and income, such as a cash transfer program. Initial evidence on programs offering both cash transfers and parenting support reveals significant effects on home learning environments and developmental outcomes (Sommer et al., 2024).

Parenting self-efficacy is another key predictor of family engagement. Although RF analyses found it important for both retention and involvement, it demonstrated a significant linear relation only with retention in regression analyses, with lower levels being associated with high retention. This finding corroborates prior studies showing that parents with more psychosocial risks are more likely to enroll and continue attending PPPs (Mauricio, Mazza, et al., 2018; Damashek et al., 2011) and aligns with previous PlayReadVIP findings (Miller et al., 2020). Considering the program’s strength-based approach, it is possible that parents with lower parenting self-efficacy felt empowered and more competent in their role after attending PlayReadVIP sessions and thus were more likely to sustain their engagement. This provides additional evidence that PlayReadVIP may be effectively targeting parents who are likely to benefit the most from receiving parenting support. The nonsignificant linear relation between parenting self-efficacy and involvement, similar to other predictors that were found to be important by RF analyses but did not show significant relations with family engagement in regression analyses, suggests that those sociodemographic (i.e., maternal age, first-time parent, Black/African American, parent education, and household income), psychosocial (i.e., emotional support and family service) factors, and neighborhood safety may be associated with family engagement in complex ways that cannot be captured by simple linear relations.

Limitations and Future Directions

Due to constraints on sample size, we were not able to examine the shape (e.g., linear or nonlinear) and direction of the relations between each important predictor and family engagement or potential interactions among predictors in the RF analyses. Future studies with larger sample sizes should utilize partial dependence plots to visualize how the important predictors are related to family engagement and then formulate corresponding regression models to test those relations (Fife & D’Onofrio, 2023; Loef et al., 2022). In addition, qualitative research is needed to unravel the mechanisms through which family factors and contextual factors outside the home influence family engagement. This could help researchers understand the specific challenges and concerns parents were facing during COVID-19 that might have hindered them from attending PlayReadVIP sessions. The reliability of the parenting self-efficacy scale was approaching acceptable and findings pertaining to this variable should be replicated in future studies.

Furthermore, although this study is among the first to examine a multitude of family and neighborhood characteristics and COVID-19 as predictors of family engagement in early childhood PPPs, it is constrained in the operationalizations of family engagement. First, it only assessed the behavioral aspects of family engagement and did not measure attitudinal or cognitive aspects, such as the degree to which PlayReadVIP aligns with parents’ beliefs about child rearing, parents’ satisfaction about the program, and perception of their relationship with the PlayReadVIP coach (Eisner & Meidert, 2011). Future studies should consider including these aspects to obtain a more comprehensive understanding of family engagement. Second, it did not measure involvement between sessions (e.g., how frequently families practiced the learned skills and/or used the PlayReadVIP toys and books at home). Although such data might be difficult to obtain, they would provide important information on the extent to which families were engaged in the program. Third, our sample size was not large enough to examine differential patterns of attendance (e.g., a family who attended the 1-, 2-, and 4-month sessions versus a family who attended the 1-, 2-, and 9-month sessions). Fourth, the operationalization of “involvement” and “partial sessions” overlapped because most of the partial sessions were due to nonparticipation in the video recording and review component. Although this study focused on involvement, future studies should explore different reasons for partial sessions to better understand the barriers to family engagement.

Lastly, given that this study was conducted primarily during the COVID-19 pandemic, findings on the predictors of family engagement need to be replicated in other contexts. Nonetheless, they may provide insights into potential strategies for engaging families who face community-level stressors.

Conclusion

This study demonstrates that family engagement in early childhood PPPs is associated with a multitude of sociodemographic, psychosocial, and contextual factors and hindered particularly by the COVID-19 pandemic. By understanding which families enroll, attend sessions continuously, and are actively involved, we can enhance implementation strategies to reach families who are traditionally underrepresented in parenting programs and more effectively retain families over time. Our findings also support the idea that strength-based, pediatric primary care PPPs that reinforce confidence in parenting can facilitate high attendance and potentially engage families with higher needs.

Supplementary Material

Supplemental Materials

Funding.

Research reported in this manuscript was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD) under Award Number R01HD096909, the Community Foundation of Greater Flint, the Ben F. Bryer Foundation, the Hagerman Foundation, the Kiwanis Club of Flint, and the Annenburg Foundation.

Footnotes

Conflicts of interest. The authors declare that they have no conflict of interest.

Ethical approval. The randomized controlled trial where data were drawn from for this study was registered with clinicaltrials.gov (NCT03945552). All study procedures and measures were reviewed and approved by the Institutional Review Board at NYU Grossman School of Medicine acting as the single IRB (s18-01347) for this study. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent. Informed consent was obtained from all individual participants included in the study.

Code/data availability.

The dataset is not readily available as the data collection for the randomized controlled trial is ongoing. Dataset will be available upon completion of primary study aims of the randomized controlled trial by request to the corresponding author. The study analysis code is available upon request to the corresponding author.

References

  1. Baker CN, Arnold DH, & Meagher S (2011). Enrollment and attendance in a parent training prevention program for conduct problems. Prevention Science, 12(2), 126–138. 10.1007/s11121-010-0187-0 [DOI] [PubMed] [Google Scholar]
  2. Barnett ER, Knight-Zhang EM, Haskell HG, Merriweather KA, Gaspar HA, Parton RR, & Jankowski MK (2023). Family engagement in services during COVID-19: A mixed-methods study of caregiver and staff perspectives. Journal of Pediatric Health Care, 37(2), 142–152. 10.1016/j.pedhc.2022.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Breslau N, Peterson EL, Kessler RC, & Schultz LR (1999). Short screening scale for DSM-IV posttraumatic stress disorder. The American Journal of Psychiatry, 156(6), 908–911. 10.1176/ajp.156.6.908 [DOI] [PubMed] [Google Scholar]
  4. Brody GH, Murry VM, Chen Y, Kogan SM, & Brown AC (2006). Effects of family risk factors on dosage and efficacy of a family-centered preventive intervention for rural African Americans. Prevention Science: The Official Journal of the Society for Prevention Research, 7(3), 281–291. 10.1007/s11121-006-0032-7 [DOI] [PubMed] [Google Scholar]
  5. Burney V, McCann CM, & Arnold-Saritepe A (2024). Parent engagement in child-focused interventions: a systematised review of qualitative allied health literature. Child & Youth Care Forum, 53(6), 1451–1486. 10.1007/s10566-024-09797-6 [DOI] [Google Scholar]
  6. Canfield CF, Miller EB, Zhang Y, Shaw D, Morris P, Galan C, & Mendelsohn AL (2023). Tiered universal and targeted early childhood interventions: Enhancing attendance across families with varying needs. Early Childhood Research Quarterly, 63, 362–369. 10.1016/j.ecresq.2023.01.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cates CB, Weisleder A, Johnson SB, Seery AM, Canfield CF, Huberman H, Dreyer BP, & Mendelsohn AL (2018). Enhancing parent talk, reading, and play in primary care: Sustained impacts of the Video Interaction Project. The Journal of Pediatrics, 199, 49–56.e1. 10.1016/j.jpeds.2018.03.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cohen S, Kamarck T, & Mermelstein R (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24(4), 385. 10.2307/2136404 [DOI] [PubMed] [Google Scholar]
  9. Conger RD, & Conger KJ (2002). Resilience in midwestern families: Selected findings from the first decade of a prospective, longitudinal study. Journal of Marriage and Family, 64(2), 361–373. 10.1111/j.1741-3737.2002.00361.x [DOI] [Google Scholar]
  10. Damashek A, Doughty D, Ware L, & Silovsky J (2011). Predictors of client engagement and attrition in home-based child maltreatment prevention services. Child Maltreatment, 16(1), 9–20. 10.1177/1077559510388507 [DOI] [PubMed] [Google Scholar]
  11. Desimpelaere EN, Soenens B, Prinzie P, Waterschoot J, Vansteenkiste M, Morbée S, Schrooyen C, & De Pauw SSW (2023). Parents’ stress, parental burnout, and parenting behavior during the covid-19 pandemic: Comparing parents of children with and without complex care needs. Journal of Child and Family Studies, 32(12), 3681–3696. 10.1007/s10826-023-02702-0 [DOI] [Google Scholar]
  12. Dumka LE, Stoerzinger HD, Jackson KM, & Roosa MW (1996). Examination of the cross-cultural and cross-language equivalence of the Parenting Self-Agency Measure. Family Relations, 45(2), 216–222. 10.2307/585293 [DOI] [Google Scholar]
  13. Eisner M, & Meidert U (2011). Stages of parental engagement in a universal parent training program. The Journal of Primary Prevention, 32(2), 83–93. 10.1007/s10935-011-0238-8 [DOI] [PubMed] [Google Scholar]
  14. Fife DA, & D’Onofrio J (2023). Common, uncommon, and novel applications of random forest in psychological research. Behavior Research Methods, 55(5), 2447–2466. 10.3758/s13428-022-01901-9 [DOI] [PubMed] [Google Scholar]
  15. Friborg O, Hjemdal O, Rosenvinge JH, & Martinussen M (2003). A new rating scale for adult resilience: What are the central protective resources behind healthy adjustment? International Journal of Methods in Psychiatric Research, 12(2), 65–76. 10.1002/mpr.143 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Garvey C, Julion W, Fogg L, Kratovil A, & Gross D (2006). Measuring participation in a prevention trial with parents of young children. Research in Nursing & Health, 29(3), 212–222. 10.1002/nur.20127 [DOI] [PubMed] [Google Scholar]
  17. Gavidia-Payne S, Denny B, Davis K, Francis A, & Jackson M (2015). Parental resilience: A neglected construct in resilience research. Clinical Psychologist, 19(3), 111–121. 10.1111/cp.12053 [DOI] [Google Scholar]
  18. Hackworth NJ, Matthews J, Westrupp EM, Nguyen C, Phan T, Scicluna A, Cann W, Bethelsen D, Bennetts SK, & Nicholson JM (2018). What influences parental engagement in early intervention? Parent, program and community predictors of enrolment, retention and involvement. Prevention Science, 19(7), 880–893. 10.1007/s11121-018-0897-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Haine-Schlagel R, & Walsh NE (2015). A review of parent participation engagement in child and family mental health treatment. Clinical Child and Family Psychology Review, 18(2), 133–150. 10.1007/s10567-015-0182-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Heinrichs N, Bertram H, Kuschel A, & Hahlweg K (2005). Parent recruitment and retention in a universal prevention program for child behavior and emotional problems: Barriers to research and program participation. Prevention Science, 6(4), 275–286. 10.1007/s11121-005-0006-1 [DOI] [PubMed] [Google Scholar]
  21. Jeong J, Franchett EE, Oliveira C. V. R. de, Rehmani K, & Yousafzai AK (2021). Parenting interventions to promote early child development in the first three years of life: A global systematic review and meta-analysis. PLOS Medicine, 18(5), e1003602. 10.1371/journal.pmed.1003602 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Katz KS, El-Mohandes A, Johnson DM, Jarrett M, Rose A, & Cober M (2001). Retention of low income mothers in a parenting intervention study. Journal of Community Health, 26(3), 203–218. 10.1023/A:1010373113060 [DOI] [PubMed] [Google Scholar]
  23. Loef B, Wong A, Janssen NAH, Strak M, Hoekstra J, Picavet HSJ, Boshuizen HCH, Verschuren WMM, & Herber G-CM (2022). Using random forest to identify longitudinal predictors of health in a 30-year cohort study. Scientific Reports, 12(1), 10372. 10.1038/s41598-022-14632-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Mathis A, Rooks R, & Kruger D (2016). Improving the neighborhood environment for urban older adults: social context and self-rated health. International Journal of Environmental Research and Public Health, 13(1), Article 1. 10.3390/ijerph13010003 [DOI] [Google Scholar]
  25. Mauricio AM, Mazza GL, Berkel C, Tein J-Y, Sandler IN, Wolchik SA, & Winslow E (2018). Attendance trajectory classes among divorced and separated mothers and fathers in the new beginnings program. Prevention Science, 19(5), 620–629. 10.1007/s11121-017-0783-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mauricio AM, Tein J-Y, Gonzales NA, Millsap RE, Dumka LE, & Berkel C (2014). Participation patterns among Mexican–American parents enrolled in a universal intervention and their association with child externalizing outcomes. American Journal of Community Psychology, 54(3–4), 370–383. 10.1007/s10464-014-9680-0 [DOI] [PubMed] [Google Scholar]
  27. Mendelsohn AL, Cates CB, Weisleder A, Berkule Johnson S, Seery AM, Canfield CF, Huberman HS, & Dreyer BP (2018). Reading aloud, play, and social-emotional development. Pediatrics, 141(5), e20173393. 10.1542/peds.2017-3393 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Miller EB, Canfield CF, Morris PA, Shaw DS, Cates CB, & Mendelsohn AL (2020). Sociodemographic and psychosocial predictors of VIP attendance in smart beginnings through 6 months: Effectively targeting at-risk mothers in early visits. Prevention Science, 21(1), 120–130. 10.1007/s11121-019-01044-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Miller EB, Canfield CF, Roby E, Wippick H, Shaw DS, Mendelsohn AL, & Morris-Perez PA (2024). Enhancing early language and literacy skills for racial/ethnic minority children with low incomes through a randomized clinical trial: The mediating role of cognitively stimulating parent–child interactions. Child Development, 95(4), 1172–1185. 10.1111/cdev.14064 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Miller EB, Roby E, Zhang Y, Coskun L, Rosas JM, Scott MA, Gutierrez J, Shaw DS, Mendelsohn AL, & Morris-Perez PA (2023). Promoting cognitive stimulation in parents across infancy and toddlerhood: A randomized clinical trial. The Journal of Pediatrics, 255, 159–165.e4. 10.1016/j.jpeds.2022.11.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Oberg C, Hodges HR, Gander S, Nathawad R, & Cutts D (2022). The impact of COVID-19 on children’s lives in the United States: Amplified inequities and a just path to recovery. Current Problems in Pediatric and Adolescent Health Care, 52(7), 101181. 10.1016/j.cppeds.2022.101181 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. O’Brien RA, Moritz P, Luckey DW, McClatchey MW, Ingoldsby EM, & Olds DL (2012). Mixed methods analysis of participant attrition in the nurse-family partnership. Prevention Science, 13(3), 219–228. 10.1007/s11121-012-0287-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Oxford M, Fleming C, & Barnard K (2015). Difficult Life Circumstances scale: A resource manual (2nd ed.). Seattle, WA: NCAST Programs. [Google Scholar]
  34. Piccolo LR, Roby E, Canfield CF, Seery AM, Weisleder A, Cates CB, Tutasig L, Matalon M, Custode A, Rodriguez L, & Mendelsohn AL (2024). Supporting responsive parenting in real-world implementation: Minimal effective dose of the Video Interaction Project. Pediatric Research, 95(5), 1295–1300. 10.1038/s41390-023-02916-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Prado G, Pantin H, Schwartz SJ, Lupei NS, & Szapocznik J (2006). Predictors of Engagement and Retention into a Parent-Centered, Ecodevelopmental HIV Preventive Intervention for Hispanic Adolescents and their Families. Journal of Pediatric Psychology, 31(9), 874–890. 10.1093/jpepsy/jsj046 [DOI] [PubMed] [Google Scholar]
  36. Prime H, Walsh F, & Masten AS (2023). Building family resilience in the wake of a global pandemic: Looking back to prepare for the future. Canadian Psychology / Psychologie Canadienne, 64(3), 200–211. 10.1037/cap0000366 [DOI] [Google Scholar]
  37. Reyno SM, & McGrath PJ (2006). Predictors of parent training efficacy for child externalizing behavior problems – a meta-analytic review. Journal of Child Psychology and Psychiatry, 47(1), 99–111. 10.1111/j.1469-7610.2005.01544.x [DOI] [PubMed] [Google Scholar]
  38. Robinson L, Adair P, Coffey M, Harris R, & Burnside G (2016). Identifying the participant characteristics that predict recruitment and retention of participants to randomised controlled trials involving children: A systematic review. Trials, 17(1), 294. 10.1186/s13063-016-1415-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Rothacher Y, & Strobl C (2024). Identifying informative predictor variables with random forests. Journal of Educational and Behavioral Statistics, 49(4), 595–629. 10.3102/10769986231193327 [DOI] [Google Scholar]
  40. Sandler IN, Schoenfelder EN, Wolchik SA, & MacKinnon DP (2011). Long-term impact of prevention programs to promote effective parenting: Lasting effects but uncertain processes. Annual Review of Psychology, 62(Volume 62, 2011), 299–329. 10.1146/annurev.psych.121208.131619 [DOI] [Google Scholar]
  41. Solís-Cordero K, Duarte LS, & Fujimori E (2022). Effectiveness of remotely delivered parenting programs on caregiver-child interaction and child development: A systematic review. Journal of Child and Family Studies, 31(11), 3026–3036. 10.1007/s10826-022-02328-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sommer TE, Franchett E, Yoshikawa H, & Lombardi J (2024). A global call for two-generation approaches to child development and caregivers’ livelihoods. Child Development Perspectives, 18(4), 204–214. 10.1111/cdep.12516 [DOI] [Google Scholar]
  43. Staudt M (2007). Treatment engagement with caregivers of at-risk children: Gaps in research and conceptualization. Journal of Child and Family Studies, 16(2), 183–196. 10.1007/s10826-006-9077-2 [DOI] [Google Scholar]
  44. Teresi JA, Ocepek-Welikson K, Kleinman M, Ramirez M, & Kim G (2016a). Measurement equivalence of the Patient Reported Outcomes Measurement Information System® (PROMIS®) Anxiety Short Forms in ethnically diverse groups. Psychological Test and Assessment Modeling, 58(1), 183–219. [PMC free article] [PubMed] [Google Scholar]
  45. Teresi JA, Ocepek-Welikson K, Kleinman M, Ramirez M, & Kim G (2016b). Psychometric properties and performance of the Patient Reported Outcomes Measurement Information System® (PROMIS®) Depression Short Forms in ethnically diverse groups. Psychological Test and Assessment Modeling, 58(1), 141–181. [PMC free article] [PubMed] [Google Scholar]
  46. U.S. Census Bureau. (2024). Poverty thresholds. https://www.census.gov/data/tables/time-series/demo/income-poverty/historical-poverty-thresholds.html
  47. U.S. Census Bureau. (2024). QuickFacts: Flint city, Michigan. https://www.census.gov/quickfacts/fact/table/flintcitymichigan/PST045224
  48. Weisleder A, Cates CB, Dreyer BP, Berkule Johnson S, Huberman HS, Seery AM, Canfield CF, & Mendelsohn AL (2016). Promotion of positive parenting and prevention of socioemotional disparities. Pediatrics, 137(2), e20153239. 10.1542/peds.2015-3239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Whittaker KA, & Cowley S (2012). An effective programme is not enough: A review of factors associated with poor attendance and engagement with parenting support programmes. Children & Society, 26(2), 138–149. 10.1111/j.1099-0860.2010.00333.x [DOI] [Google Scholar]
  50. Zhang L, Ssewanyana D, Martin M-C, Lye S, Moran G, Abubakar A, Marfo K, Marangu J, Proulx K, & Malti T (2021). Supporting child development through parenting interventions in low- to middle-income countries: An updated systematic review. Frontiers in Public Health, 9. 10.3389/fpubh.2021.671988 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Materials
NIHMS2135651-supplement-Supplemental_Materials.docx (19.1KB, docx)

Data Availability Statement

The dataset is not readily available as the data collection for the randomized controlled trial is ongoing. Dataset will be available upon completion of primary study aims of the randomized controlled trial by request to the corresponding author. The study analysis code is available upon request to the corresponding author.

RESOURCES